Abstract:Stochastic differential equation (SDE) models are the foundation for pricing and hedging financial derivatives. The drift and volatility functions in SDE models are typically chosen to be algebraic functions with a small number (less than 5) parameters which can be calibrated to market data. A more flexible approach is to use neural networks to model the drift and volatility functions, which provides more degrees-of-freedom to match observed market data. Training of models requires optimizing over an SDE, which is computationally challenging. For European options, we develop a fast stochastic gradient descent (SGD) algorithm for training the neural network-SDE model. Our SGD algorithm uses two independent SDE paths to obtain an unbiased estimate of the direction of steepest descent. For American options, we optimize over the corresponding Kolmogorov partial differential equation (PDE). The neural network appears as coefficient functions in the PDE. Models are trained on large datasets (many contracts), requiring either large simulations (many Monte Carlo samples for the stock price paths) or large numbers of PDEs (a PDE must be solved for each contract). Numerical results are presented for real market data including S&P 500 index options, S&P 100 index options, and single-stock American options. The neural-network-based SDE models are compared against the Black-Scholes model, the Dupire's local volatility model, and the Heston model. Models are evaluated in terms of how accurate they are at pricing out-of-sample financial derivatives, which is a core task in derivative pricing at financial institutions.

Machine learning methods for American-style path-dependent contracts

Machine Learning Methods for Pricing Financial Derivatives

Pricing American Options using Machine Learning Algorithms

Parallel American Monte Carlo

Monte Carlo Methods and Path-Generation techniques for Pricing Multi-asset Path-dependent Options

Pricing and hedging American-style options with deep learning

Deep learning of optimal exercise boundaries for American options

American options in time-dependent one-factor models: Semi-analytic pricing, numerical methods and ML support

A deep learning method for pricing high-dimensional American-style options via state-space partition

Prediction-Enhanced Monte Carlo: A Machine Learning View on Control Variate

Learning Bermudans

American Call Options Pricing With Modular Neural Networks

Jump Diffusion-Informed Neural Networks with Transfer Learning for Accurate American Option Pricing under Data Scarcity

Machine learning for quantitative finance: fast derivative pricing, hedging and fitting

Non-Linear pricing with differential machine learning

A Gaussian Process Based Method with Deep Kernel Learning for Pricing High-dimensional American Options

Prediction of Barrier Option Price Based on Antithetic Monte Carlo and Machine Learning Methods

Deep Joint Learning valuation of Bermudan Swaptions

On the pricing of capped volatility swaps using machine learning techniques

Pricing European Options with Google AutoML, TensorFlow, and XGBoost

American option pricing using generalised stochastic hybrid systems